OR in logistic regression

[MANOJ KUMAR]

Dear all 

Hope you are all doing great..!! 

1. I have a question regarding biomarker prediction and diagnosis studies while performing ROC analysis, t-test, logistic regression, quantile regression, and LASSO regression. I have seen some articles using a combination of these methods, while others use only two or three. What is the correct approach to selecting the appropriate method?

2. In my study, I am predicting a non-healing biomarker, but my odds ratio is 0.567. Since this is less than one, how should I interpret this result? can i use the 1-X(1- Biomarker)

3.  In a clinical trial, chronic wound patients are recruited, and initially, all are categorized as non-healing. However, at a later visit (V3), we identify both healing and non-healing patients. This is a prospective study design plz correct me. How should I perform sample size calculation when I have an odds ratio and ROC analysis results?

Would you like me to provide answers to these questions as well?

with regards, Dr. Manoj Kumar

Postdoctoral Associate McGowan Institute for Regenerative Medicine (MIRM) Department of Surgery University of Pittsburgh 412 Bridgeside Point II, 450 Technology Dr, Pittsburgh, PA - 15219 Phone: +1-4125898944# Email mak...@pitt.edu & manoj...@jnu.ac.in

https://www.nature.com/articles/s41598-024-74426-0